Abstract Blood is a vital and limited resource in the brain: neuronal activity requires supplies of oxygen and glucose, and deprivation of flow to even restricted regions leads inevitably to cell death. How is the distribution of blood controlled relative to the changing needs of cortex? The answers bear directly on cortical function. Developmentally, they yield design rules for vascular architecture. From the perspective of physiology, the answers define state variables that relate neuronal activity to changes in blood flow. This has implications for neuroimaging, as the extraction of oxygen from blood is the basis of contrast in BOLD fMRI. Lastly, for neurology, the answers determine the resistance of cognitive processes to vascular trauma and disease. The current literature on molecular signaling between neurons and arterioles, based mainly on slice preparations, points to a competition between vasoactive signaling molecules that dilate arteriole smooth muscle and those that cause constriction. Mixtures of such signaling molecules are released by inhibitory interneurons, which directly contact muscle, and by excitatory neurons, which act via the excitation of astrocytes that encapsulate vessels. This picture is compelling yet incomplete, as natural stimuli change flow on the subsecond level while the literature points to changes on tens of seconds. The main challenge is to assay both signaling molecules and flow in vivo. We will meet this challenge by combining our strengths in in vivo electrophysiology and imaging with our methods for precise quantification of blood flow and newly engineered indicators for signaling molecules. We envision a model of neurovascular control that maps the activity of different neuronal subtypes to changes in vascular tone. The dynamics of the underlying signaling molecules form the state variables, much as channel dynamics form the state variables in single neuron dynamics. Such a model will clarify neurovascular disease models and neuroimaging studies.